Short-term particulate matter contamination severely compromises insect antennal olfactory perception

Sub Levels

Documentation of PM on insect antennae

Wild houseflies (Musca domestica L.) were captured between June and July 2020, when Air Quality Index (AQI) conditions in Beijing were defined as low (AQI ≤ 50), moderate (50 <AQI ≤ 100), and high (100 <AQI ≤ 150), according to US-EPA 2016 standards51. The ambient AQI measures in Beijing were obtained from The World Air Quality Project52 (Fig. S1b, c), and the indoor and outdoor AQI were confirmed by a Hanvon PM detector M1 (Hanwang Technology Co., Ltd, Beijing, China). The houseflies were carefully pinned and air-dried individually in a lab equipped with air purifier (Samsung AX70J7000WT Samsung Co., Ltd, Gyeonggi-do, South Korea) to prevent cross contamination before preparation for SEM (Scanning Electron Microscope) images.

We sampled PM from the polluted air using a glass fibre filter fixed in a Th-150c medium flow Atmospheric Particulate Matter Sampler, paired with a PM-100 multi-stage Particulate Matter Collector (Wuhan Tianhong Environmental Protection Industry Co., Ltd, Wuhan, China). Samples were obtained from the campus of Beijing Forestry University for 2 h during the period of insect exposure (see below) with a flow of 100 L/min. Clean, glass fibre filters were prepared as controls.

Housefly pupae were sourced from the laboratory population maintained at the National Institute for Communicable Disease Control and Prevention, Chinese Centre for Disease Control and Prevention. The insects were cultured at 25 ± 1 °C with relative humidity of 30–50%, and 10:14 h light: dark photoperiod; the larvae were reared on a diet of water, wheat bran and milk powder mixture at the mass ratio of 200:100:1; and adults fed with aqueous 10% sucrose solution ad libitum. Pupae and adult flies were housed in a laboratory fitted with an air purifier (Samsung AX70J7000WT Samsung Co., Ltd, Gyeonggi-do, South Korea) that maintains both the PM2.5 and PM10 concentration below 5 µg/m3 (AQI < 20). We removed the deodorisation component of the air filter, thereby ensuring, as far as possible, that gaseous pollutants (other than PM) in the laboratory environment were similar to that outside the building. Male and female adults were separated shortly after eclosion, randomly allocated to treatments, and maintained in mesh cages (35 cm × 35 cm × 30 cm) until the experimental trials.

We controlled for variation in the length of exposure to air pollution by creating two experimental groups of flies: those that had been exposed to ambient air pollution (contaminated) and those that had not (uncontaminated control). Houseflies, <2 days post eclosion, were sexed and randomly allocated to identical flight cages, each containing about 150 individuals, which were then randomly assigned to the contaminated and control treatments. All the experimental behavioural assays were conducted between March and November when houseflies are active in Beijing. Contaminated flies were exposed for 12 h to the outside ambient air in treatment laboratories with open windows in various locations in Beijing (Table S1), on days with high AQI levels (101-150), ambient temperatures at 15–25 °C, and between 7 am to 7 pm when the flies are most active. The exposure time was determined as our pilot study indicating that 12 h exposure yielded PM densities that were broadly similar to those obtained from wild caught houseflies. Control houseflies were transferred to a separate laboratory located next to the treatment laboratory with the similar size, orientation, temperature, and light conditions as the treatment laboratory, but fitted with an air filter that maintains an AQI < 20. The allocation of laboratory spaces to treatment or control was assigned randomly for each collection yielding a total of 35 batches (pairs of cages, see Table S1).

The houseflies were supplied with sufficient water and food (except for the food odour behavioural assays, see below). The flight cages were made from clean cardboard boxes of 80 cm by 40 cm by 60 cm, with five sides replaced by mosquito nets that allows sufficient aeration and activation room for the flies. To avoid any possible cross contamination, each flight cage was used once only. This procedure was conducted for batches (pairs of cages) of newly emerged flies of each generation (See Table S1), and the flies were sampled within 24 h for subsequent tests (unless specified otherwise).

We collected 25 houseflies from each of the laboratory control and contaminated groups (see above), and wild flies during light, moderate and high pollution level to compare the density of PM on the antennal surface using SEM (n = 25 each). Wild flies were collected from the campus of Beijing Forestry University at high pollution levels at the same time as the contaminated laboratory flies. The wild flies collected during the light and moderate pollution levels were collected at the same place but different pollution episodes during the same season. The two antennae of each individual were carefully mounted on the SEM stubs with one the anterior side up, and the other with the posterior side up. We additionally obtained the head, thorax, abdomen, a front leg, and a wing of each of five individuals from the contaminated laboratory group. Finally, we prepared a sample (5 mm × 5 mm) from the centre of the clean and the contaminated glass fibre filters (see above). All samples were mounted on aluminium stubs with double-sided adhesive tape, placed in a glass desiccator for 48 hours to dry thoroughly. The specimens were coated with gold in a spatter coater for 45 s (estimated gold thickness of 5 nm) and imaged on a Hitachi SU8010 field-emission SEM (Hitachi Corp., Tokyo, Japan) with the acceleration voltage of 5 kV, captured using a secondary electron detector (SE), and the scan speed of 40 S per frame.

SEM images of each antenna were stitched in image J (Fiji) using the “Tile” function from four smaller overlapping images (see Fig. 1b). The density of PM on each antenna were calculated by counting visible PM on the antennal surface, defined by the image, and divided by the area of each antenna. We counted the PM on both sides of the antennae of each individual. The density of PM on the glass fibre filters were counted by taking ten 150 µm by 150 µm quadrats (22500 µm2) on each filter. We used a total of 120 samples to calculate the density of PM (n = 15). All the PM were then classified by size (>PM10, PM10, and PM2.5)6 and the density standardised to (numbers)/mm2. PM < 0.1 µm were too small and impractical to separate from the antennal sensilla surface features and thus were not included in this study. The average densities of PM on other body parts (eye, mouthpart, thorax, abdomen, leg, wing) of the contaminated and control groups of houseflies were calculated by counting all PM within four 150 µm by 150 µm quadrats taken from each of three to four individuals (each representing 22500 µm2) haphazardly placed within each image.

To conduct the point elemental analyses on the PM accumulated on the antennae and glass fibre filters, we used the Energy Dispersive X-Ray Spectroscopy (EDX) paired with a Hitachi SU8010 SEM. A series of common elements identified in PM in previous research (C, N, Na, Mg, Al, Si, P, S, Cl, K, Ca, Ti, Cr, Fe, Cu, Zn) were selected22, and the voltage was set at 20 kV. Spot analyses were conducted on at least 100 particles randomly selected on the antennal surface of 50 individuals and glass fibre filters, and the blank area of each surface was measured to provide a baseline (Fig. S3).

Behavioural assays

We used a Y-maze olfactometer to investigate the impact of PM contamination on the response of houseflies to food and conspecific female odours (Fig. S8). The Y-maze comprised three acrylic tubes (internal diameter 20 mm; length 50 mm) connected at 120°. The two arms were connected to clean odourless Teflon tubes (internal diameter 8 mm) that delivered either the odour stimulus or clean air control. The air was filtered with an activated carbon filter and an air scrubber, then pumped into the olfactometer at 0.28 m/s at the arm entrance through a long tube to maintain a laminar flow. After each trial, the Y-maze was cleaned with ethanol and detergent, and the position of odour stimulus and solution control were rotated to remove positional and arm effects for each trial. All trials were conducted at 20–25 °C, 40–60% humidity, with both arms of the Y-maze exposed to similar light intensity. Pilot experiments confirmed that the flies showed no preference when presented with the same odour at both arms.

We examined the response of houseflies to food (honey solution) and conspecific female odours (cuticle extraction of female flies, following Silhacek et al.53). Female cuticle extraction stock was prepared by soaking twenty 3–5 days old virgin female flies in 10 ml hexane for 1 min. We diluted the honey solution to 1, 2, 5, 8, and 10 times with water, and the sex attractant to 2, 10, 40, 70, and 100 times with n-Hexane. The odour treatment was delivered by applying 25 μL attractant solution to a piece of 10 mm × 10 mm filter paper, which was allowed to dry completely for at least 15 min. Twenty-five microlitre pure solvent (water for food odour assays, n-Hexane for female odours) applied onto a piece of filter paper was used as a control, which was also dried naturally for at least 15 min. Male and female houseflies were fed with water only before the food attractant assays, and only males were used for the sex attractant assays (females did not show any preference to this odour in preliminary analyses). Sixty males and/or females were tested on each dilution level of all experimental comparisons, yielding 600 trials for each assay. Each housefly was tested once only.

The test housefly was chilled at 4 °C for 60 s to reduce its activity level before being gently transferred to a cylindrical tube (35 mm long, 20 mm diameter, with mesh at one end) that was then inserted into the exit arm of the Y-maze. The trials were commenced when the recovered housefly entered the Y-maze olfactometer, and we recorded which arm of the Y-maze the housefly first entered, moving beyond 10 mm from the entrance to the arm (delimitated by a line on the tube). Each trial lasted for 2 minutes with the observer being blind to the fly treatment, and we discarded trials in which the individual did not enter either of the arms of the Y-maze.


Glass capillary tubes (outer diameter 2 mm) were pulled into micropipettes using a Dual-Stage Glass Micropipette Puller (PC-100, Narishige, Tokyo, Japan). The tips of the glass micropipette were polished to ensure the inner diameter of the glass electrode was slightly larger than the diameter of the housefly antenna. An Ag-AgCl electrode was placed in the glass micropipette that was filled with Ringer buffer (SL6438, Coolaber, Beijing, China) to record the antennal responses. The legs and wings of the chilled housefly were removed, and the individual was immobilised in a plastic micro pipette tip, with only the head exposed54. The two electrodes were positioned using micromanipulators under a stereoscope, with the indifferent electrode inserted into a compound eye and the recording electrode connected to an antenna, with the tip removed to form a complete circuit. EAGs were analysed and stored on the computer using EAGPro software (Syntech, Hilversum, Netherlands). The procedure was deemed successful if a relatively stable baseline was observed on the recording equipment. The odour stimulations were delivered at the air speed of 0.5 m/s and duration of 0.5 s.

The odour was delivered through a steel nozzle (diameter: 6 mm; length: 15 cm) positioned 2 cm from the antenna (Fig. S9a). The purified and humidified air generated by a stimulus flow controller (CS-55, Syntech, Buchen-bach, Germany) was blown over the antenna at 12.5 mL/s. The stimulus flow controller generated air pulses through the odour cartridge at a flow rate of 10 mL/s, a compensating air flow was provided to maintain a constant current. We recorded the antennal response of 15 control and 15 contaminated houseflies (contaminated within two days of eclosion) of each sex to different concentrations of honey solution (n = 15) and 97% sex pheromone (Z)-9-Tricosene (Sigma-Aldrich)55 (n = 15). Aged flies were maintained in the laboratory environment with the conditions described above for seven to ten days before EAG assays (n = 20). Similar rates of survival were observed between treatments, as over 90% of the flies in both groups survived. The honey was diluted with water to 2, 5, 8, 10, 20, 50, and 100 times (v/v), and the sex attractant was diluted with hexane to 10, 20, 30, 50, 70, and 100 times (v/v). Each housefly was tested against every dilution level of the stimulus. The baseline action potential and the response to solvent were also recorded for each housefly. Filter paper strips (5 mm×50 mm), adsorbed with 25 μL of the attractant solution and left until the solvent is dried completely for 3 min, was inserted into a glass Pasteur pipette to introduce the stimuli into the airstream56. Pasteur pipette and filter paper strips were used once only. Filter paper loaded with solvent was used as a control before and after presenting the stimuli, providing a baseline of each antenna. The stimuli were added to the air stream in increasing concentration at an interval of at least 30 s, to avoid sensory adaptation during recordings.

The recorded signals were amplified by an IDAC interface amplifier (IDAC-4, Syntech, Buchenbach, Germany), and the data analysed using Autospike 3.4 (Syntech, Buchenbach, Germany). We recorded the maximum antennal response of each trial, determined by the software EagPro (Fig. S9c). The responses were standardised as a %, using the formula: \(\frac{100\times \left({measured\; value}-{solvent\; response\; value}\right)}{{solvent\; response\; value}}\).

Transcriptome analysis and RT-qPCR

Adult male and female houseflies of the same age (48 h post eclosion) from both control and contaminated treatments were exposed to PM using the same protocol as the previous experiments in both spring and summer. Within 24 h after the treatments, the flies were removed from their cages and immediately frozen in liquid nitrogen for 5 min, and then stored at −80 °C. The antennae were excised (under a stereoscope) from the rest of the body using sterilized tweezers and placed in separate, sterilized cryopreservation tubes. Dry ice was used to ensure the temperature of equipment was maintained as low as possible. Each sample comprised 60 houseflies, and we obtained three biological replicates for each gender and treatment, yielding a total of 24 samples per season (three biological replicas × two body parts × two sexes × two treatments). The samples were stored at −80 °C until sequencing.

The total RNA was extracted using the TRIzol method57. RNA sequencing (RNAseq) was completed by Allwegene Technology Inc., Beijing. The cDNA library was then constructed using polymerase chain reaction (PCR) amplification. RNA-seq was performed with the PE150 sequencing strategy using an Illumina second-generation high-throughput sequencing platform. RNA-seq reads with inferior quality or adaptors were filtered. Clean read data were processed using Tophat2 and Cufflinks software to complete the alignment of transcriptomes. Differentially expressed genes and transcripts were then filtered for false discovery rate (FDR) adjusted P  0.05. The reference whole genome library and annotation follows that of Scott et al.58.

We assessed whether the impact of PM is associated with any changes in gene expression level by comparing genes that are associated with various physiological processes expressed in antennae and the rest of the body. The common up- and downregulated mRNAs were highlighted with volcano plots. We subsequently used GOseq59 in gene ontology (GO) enrichment analysis to identify their functions, using the Kyoto Encyclopaedia of Genes and Genomes (KEGG). Volcano plots were generated using GraphPad Prism 8.0 (GraphPad Software, La Jolla California USA).

We selected 25 relatively highly expressed DEGs (differentially expressed genes) critical for the function of fly antennae, using RT-qPCR to validate the transcriptomic analysis results. Total RNA was extracted from three biological replica of housefly antenna and body samples, each containing 80 individuals, yielding 24 samples (three biological replica × two body parts × two sexes × two treatments). The first-strand cDNA was synthesised from 1 μg of total RNA extracted using PrimeScript™ RT reagent Kit with gDNA Eraser. RT-qPCR reaction contained 1 μl cDNA (100 ng), of 0.2 μl each primer (10 μmol), 5 μl 2x SYBR Green PCR buffer (Bio-Rad, Hercules, CA, USA) and 3.6 μl nuclease-free water. Primer pairs used for each gene are listed in Supplementary Table S2. The PCR conditions for the amplification action were as follows: 3 min at 95 °C, followed by 40 cycles of 10 s at 95 °C, 60 s at 54 °C, and 30 s at 60 °C. The PCR products were examined and calculated using the 2-∆∆CT method. The relative expression was calculated against that of the house keeping gene GAPDH.

Statistics and reproducibility

All statistical analyses were performed using JMP 14 pro (SAS, USA). For SEM data, we used generalised linear mixed models with Tukey post hoc test to investigate variation in the density of PM between different body parts of contaminated flies, and on the antennae between contaminated and wild flies, with individual identity included as a random effect (Table S3, S4). The proportion of each class of PM size on the antennae of the contaminated group and the glass fibre filters were compared using Wilcoxon test (Table S5). For behavioural assays, we used mixed models with a binomial distribution and logit link function to analyse the response of houseflies, with treatment (contaminated, not contaminated), dilution levels and their interaction as fixed effects and fly collection batch identity as a random effect (Table S6). For EAG data, we used mixed models with Tukey post hoc test to explain the variation in relation to blank value, with treatment (contaminated or uncontaminated) and dilution as fixed effects and fly collection batch and individual identity as a random effect (Table S7). We checked the distributions for normality of our variables before analysis. All statistical analyses employed two-tailed tests, with statistical significance set at <0.05, Bonferroni adjustment was made for multiple comparisons when it applies. Results from representative experiments are from at least 10 independently acquired individuals (Figs. S2, S3a-j, and S6).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Source link

Leave a Reply

Your email address will not be published. Required fields are marked *